Satellite-monitoring vehicle high-precision cooperative emergency method and system

Abstract

The application discloses a satellite-monitoring vehicle high-precision cooperative emergency method and system. The satellite remote sensing monitoring system constructs a water quality inversion empirical model based on historical data and monitors and alarms in real time. After receiving the alarm information, the intelligent remote sensing monitoring vehicle controls the unmanned aerial vehicle to collect hyperspectral images and the unmanned ship to monitor water quality parameters in situ through unmanned aerial vehicle sampling area and unmanned ship collection point accurate planning algorithm. The data processing system constructs a ground hyperspectral inversion model based on measured data, obtains high-precision water quality parameter layer data of the sampling area, and uses a scale elimination algorithm and a weighted least square method to dynamically correct the satellite inversion model on site. The corrected model is applied to satellite data to obtain high-precision water quality parameters, generate a water quality report and realize real-time linkage between the site and the command center through a decision command system. The application solves the problems of insufficient traditional monitoring precision and response lag, and realizes high-precision and high-efficiency cooperation of water quality monitoring.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring technology, specifically a high-precision collaborative emergency response method and system using a satellite and a monitoring vehicle. Background Technology

[0002] Traditional water quality monitoring methods mainly rely on manual sampling and fixed monitoring stations, but both have significant shortcomings in responding to sudden water pollution events: manual sampling requires a lot of manpower, material resources and time, and the monitoring range is limited, making it difficult to achieve real-time dynamic response to large areas of water; fixed monitoring stations are sparsely distributed, with a large number of monitoring blind spots, making it impossible to capture pollution signals in the first instance, and also difficult to flexibly adjust the monitoring layout according to the pollution diffusion path.

[0003] Satellite remote sensing technology, due to its wide coverage and timely monitoring capabilities, has gradually become an important means of water quality monitoring. Water quality parameter inversion based on satellite remote sensing typically relies on historical and long-term ground-based measured data, combined with synchronous satellite imagery spectral data, and constructed using empirical models. These empirical models demonstrate good robustness and applicability in monitoring water quality anomalies and can effectively reflect the trends of water quality parameters. However, the meteorological conditions of the day (such as weather changes and aerosol concentrations) can significantly affect the accuracy of satellite remote sensing data inversion, making it difficult for the water quality parameters inverted from empirical models built on historical data to meet the needs of accurate monitoring and assessment, thereby impacting decision support for relevant departments.

[0004] With the rapid development of unmanned technology, drones and unmanned vessels are increasingly being used in water quality monitoring. Drones, with their high mobility and wide monitoring range, can quickly reach target water areas to acquire high-resolution image data and some water quality parameters; while unmanned vessels can navigate on the water surface, carrying water quality monitoring equipment to achieve in-situ monitoring of water bodies. Combining the two can provide drones with measured water sample data, thereby more accurately determining the drone's water quality inversion model. However, the flight time and coverage of drones are often limited, and when facing large lakes and reservoirs, they cannot comprehensively reflect the water quality status of the entire body. Currently, there is no effective means to systematically integrate and collaboratively apply the large-scale monitoring capabilities of satellite remote sensing with the precise on-site monitoring capabilities of drones and unmanned vessels. Summary of the Invention

[0005] The purpose of this invention is to provide a high-precision collaborative emergency response method and system for satellite-monitoring vehicles, integrating the core advantages of satellite remote sensing, drones, and unmanned surface vessels (USVs). This aims to comprehensively improve the efficiency, accuracy, timeliness, and coverage of water quality monitoring. When the satellite remote sensing monitoring system issues a water quality anomaly alarm, the intelligent remote sensing monitoring vehicle (equipped with drones and USVs) will be urgently dispatched to the abnormal area to conduct on-site measurements, acquiring real-time water quality parameters and hyperspectral data for the day. Based on this measured data, on-site correction of the satellite inversion model can be achieved, significantly improving the accuracy of the satellite remote sensing inversion layer, enhancing its professionalism and practical value, and ensuring that relevant departments can obtain accurate and reliable water quality parameter data, providing a scientific basis for environmental management and decision-making.

[0006] A high-precision collaborative emergency response system integrating a satellite and a monitoring vehicle includes a central control platform integrated into an intelligent remote sensing monitoring vehicle, a satellite remote sensing monitoring system communicatively connected to the central control platform, a drone and unmanned surface vessel (USV) control system, and a data processing system. The satellite remote sensing monitoring system is used to construct a water quality parameter inversion empirical model based on historical ground-measured water quality parameter data, ground water body remote sensing reflectance data, and synchronous satellite image spectral data of the target water area. It then uses this model to invert and obtain a water quality inversion layer for the entire target water area. When the water quality parameter in a certain area of ​​the inversion layer reaches or exceeds a preset alarm threshold, an alarm message is generated and sent. The central control platform receives the alarm message and determines the drone monitoring area and USV data collection points based on the alarm message, thereby generating a flight path plan. The drone and USV control system controls the drone and USV according to the flight path plan. The UAV collects hyperspectral image data of the sampling area and transmits the collected hyperspectral image data back to the data processing system in real time. Based on the planned flight path, the unmanned surface vessel (USV) is controlled to reach a preset collection point for in-situ monitoring of water quality parameters and transmits the measured water quality parameter data back to the data processing system in real time. The data processing system receives the hyperspectral image data and the measured water quality parameter data, constructs a ground-based hyperspectral inversion model, obtains high-precision water quality parameter layer data of the sampling area, and dynamically corrects the water quality parameter inversion empirical model based on the high-precision water quality parameter layer data and satellite spectral data to generate a corrected satellite inversion model. The corrected satellite inversion model is applied to the satellite remote sensing data of the day to invert high-precision water quality parameter data, and a comprehensive water quality analysis is performed by combining the UAV hyperspectral data and the USV measured data to generate a water quality report.

[0007] Furthermore, the central control platform determines the UAV monitoring area and the UAV data collection point based on the alarm information, specifically including: S201, obtaining the alarm area area A. alert S202. The area B of a single drone shot and the preset coverage ratio P; S203. Based on the alarm area area A alertCalculate the alarm area A to be covered based on the preset coverage ratio P. coverage =A alert *P;S203, Based on the alarm area A to be covered. coverage Given the area B of a single drone shot, determine the number of sampling areas N = ceil(A) coverage / B), where ceil represents rounding up; S204. Determine the side length of a single sampling area based on the area B of the drone's single-shot range. S205. Generate an outer rectangle based on the four boundary coordinates of the alarm area, and determine the width W and height H of the outer rectangle; S206. Divide the outer rectangle into a grid based on the side length s, and calculate the number of sampling areas that can be divided in each row n = W / s and the number of sampling areas that can be divided in each column m = H / s; S207. Generate a set of potential sampling area locations C using a grid system. potential = {(j * s, i * s) | i=0,1,...,m-1; j=0,1,...,n-1}, ensuring that the potential sampling areas are uniformly distributed within the bounding rectangle; S208, based on the number of sampling areas N and the total number of potential locations n*m, calculate the uniform selection step size step = (n*m) / N, and uniformly select indices from the potential sampling area location set according to the step size step to obtain the UAV monitoring area set C. drone = {C potential [k]|k = 0,step, 2*step,...};S209、For each selected sampling area, determine its center location as the unmanned vessel sampling point P. boat ,k =(C drone,k,x +(s / 2),C drone,k,y +(s / 2)), where k = 0,1,…,N-1.

[0008] Furthermore, the data processing system dynamically corrects the water quality parameter inversion empirical model based on the high-precision water quality parameter layer data and satellite spectral data to generate a corrected satellite inversion model. Specifically, this includes: S601, using a scale elimination algorithm, resampling the high-precision water quality parameter layer data of the sampling area obtained in step S5 to a resolution consistent with the satellite data to obtain satellite-scale water quality parameter data; S602, matching the satellite-scale water quality parameter data with the satellite spectral data of the corresponding area to construct a correction dataset; S603, using a weighted least squares method, with the satellite-scale water quality parameter data in the correction dataset as the true label, iteratively optimizing the parameters of the water quality parameter inversion empirical model constructed in step S1 until a preset error threshold is met to obtain the corrected satellite inversion model.

[0009] Furthermore, the scale elimination algorithm includes: calculating a spatial weighting factor based on the Euclidean distance between the sub-meter pixel and the satellite pixel center. ; Spectral similarity weighting factor is calculated based on the cosine similarity between sub-meter pixel hyperspectral reflectance and average satellite pixel hyperspectral reflectance. Spatial weighting factor Weighting factor for spectral similarity After normalization and weighted fusion, a comprehensive weight factor is obtained. According to the comprehensive weighting factor The water quality parameter values ​​of the satellite pixels are obtained by weighted summation of the sub-meter level pixel values. .

[0010] Furthermore, the objective function of the weighted least squares method is: ; ; .

[0011] in, Let be the objective function. To correct the total number of samples in the dataset; True label value of satellite pixels Confidence weights; This is the mathematical expression of the optimal satellite inversion model; Let be the spectral feature vector of the j-th satellite pixel; These are the actual labeled values ​​of water quality parameters at the satellite scale. For drones with sub-meter pixel resolution Water quality parameters The measured confidence level is calculated based on the in-situ measured data of the unmanned vessel and the inverted data of the UAV. For drones with sub-meter pixel resolution The water quality parameters were monitored in situ by the unmanned surface vessel that was closest to the water in the water. For drones with sub-meter pixel resolution The water quality parameters were retrieved from the ground-based hyperspectral retrieval model.

[0012] A high-precision collaborative emergency response method using a satellite-monitoring vehicle, employing the system described above, includes the following steps: S1, the satellite remote sensing monitoring system constructs a water quality parameter inversion empirical model based on historical ground-measured water quality parameter data, ground water body remote sensing reflectance data, and synchronous satellite image spectral data of the target water area. The system then uses this model to obtain a water quality inversion layer for the entire target water area. When the water quality parameter in a certain area of ​​the inversion layer reaches or exceeds a preset alarm threshold, an alarm message is generated and sent. S2, the central control platform of the intelligent remote sensing monitoring vehicle receives the alarm message. Based on the alarm area range and the single-shot range of the UAV, it determines the UAV monitoring area and the unmanned surface vessel (USV) collection point, generating a flight path plan, which is then sent to the UAV and USV control systems respectively. S3, the UAV and USV control systems control the UAV to monitor the sampling area according to the flight path plan. The process involves: S4) Acquiring hyperspectral image data and transmitting it back to the data processing system in real time; controlling an unmanned surface vessel (USV) to reach a preset collection point for in-situ monitoring of water quality parameters and transmitting the measured water quality parameter data back to the data processing system in real time; S5) Receiving the hyperspectral image data and measured water quality parameter data, the data processing system constructs a ground-based hyperspectral inversion model to obtain high-precision water quality parameter layer data for the sampling area; S6) Based on the high-precision water quality parameter layer data and satellite spectral data obtained in step S4, the data processing system dynamically corrects the water quality parameter inversion empirical model constructed in step S1 to generate a corrected satellite inversion model; S7) Applying the corrected satellite inversion model to the satellite remote sensing data of the day to invert high-precision water quality parameter data, and combining it with the USV hyperspectral data and USV measured data for comprehensive water quality analysis to generate a water quality report.

[0013] Furthermore, step S2, which involves determining the UAV monitoring area and the UAV data collection point based on the alarm information, specifically includes: S201, obtaining the alarm area area A. alert S202. The area B of a single drone shot and the preset coverage ratio P; S203. Based on the alarm area area A alert Calculate the alarm area A to be covered based on the preset coverage ratio P. coverage =A alert *P;S203, Based on the alarm area A to be covered. coverage Given the area B of a single drone shot, determine the number of sampling areas N = ceil(A) coverage / B), where ceil represents rounding up; S204. Determine the side length of a single sampling area based on the area B of the drone's single-shot range. S205. Generate an outer rectangle based on the four boundary coordinates of the alarm area, and determine the width W and height H of the outer rectangle; S206. Divide the outer rectangle into a grid based on the side length s, and calculate the number of sampling areas that can be divided in each row n = W / s and the number of sampling areas that can be divided in each column m = H / s; S207. Generate a set of potential sampling area locations C using a grid system. potential = {(j * s, i * s) | i=0,1,...,m-1; j=0,1,...,n-1}, ensuring that the potential sampling areas are uniformly distributed within the bounding rectangle; S208, based on the number of sampling areas N and the total number of potential locations n*m, calculate the uniform selection step size step = (n*m) / N, and uniformly select indices from the potential sampling area location set according to the step size step to obtain the UAV monitoring area set C. drone = {C potential [k]|k = 0,step, 2*step,...};S209、For each selected sampling area, determine its center location as the unmanned vessel sampling point P. boat ,k =(C drone,k,x +(s / 2),C drone,k,y +(s / 2)), where k = 0,1,…,N-1.

[0014] Further, step S6 specifically includes: S601, using a scale elimination algorithm, resampling the high-precision water quality parameter layer data of the sampling area obtained in step S5 to the same resolution as the satellite data to obtain satellite-scale water quality parameter data; S602, matching the satellite-scale water quality parameter data with the satellite spectral data of the corresponding area to construct a calibration dataset; S603, using a weighted least squares method, taking the satellite-scale water quality parameter data in the calibration dataset as the true label, iteratively optimizing the parameters of the water quality parameter inversion empirical model constructed in step S1 until a preset error threshold is met to obtain the calibrated satellite inversion model.

[0015] Furthermore, the scale elimination algorithm in step S601 includes: calculating a spatial weighting factor based on the Euclidean distance between the sub-meter pixel and the satellite pixel center. ; Spectral similarity weighting factor is calculated based on the cosine similarity between sub-meter pixel hyperspectral reflectance and average satellite pixel hyperspectral reflectance. Spatial weighting factor Weighting factor for spectral similarity After normalization and weighted fusion, a comprehensive weight factor is obtained. According to the comprehensive weighting factor The water quality parameter values ​​of the satellite pixels are obtained by weighted summation of the sub-meter level pixel values. .

[0016] Furthermore, the objective function of the weighted least squares method is: ; ; .

[0017] in, Let be the objective function. To correct the total number of samples in the dataset; True label value of satellite pixels Confidence weights; This is the mathematical expression of the optimal satellite inversion model; Let be the spectral feature vector of the j-th satellite pixel; These are the actual labeled values ​​of water quality parameters at the satellite scale. For drones with sub-meter pixel resolution Water quality parameters The measured confidence level is calculated based on the in-situ measured data of the unmanned vessel and the inverted data of the UAV. For drones with sub-meter pixel resolution The water quality parameters were monitored in situ by the unmanned surface vessel that was closest to the water in the water. For drones with sub-meter pixel resolution The water quality parameters were retrieved from the ground-based hyperspectral retrieval model.

[0018] Compared with the prior art, the present invention has the following characteristics and effects.

[0019] 1. Constructing a three-in-one collaborative monitoring system integrating satellite, drone, and ship to break through the limitations of traditional monitoring: This invention uses an intelligent remote sensing monitoring vehicle as the core carrier to integrate and coordinate the wide-area monitoring capabilities of satellite remote sensing, the high mobility and high-resolution imaging capabilities of drones, and the in-situ precision monitoring capabilities of unmanned ships. This solves the pain points of low efficiency of traditional manual sampling, limited coverage of fixed stations, and insufficient accuracy of single technical means, and achieves a comprehensive improvement in the efficiency, accuracy, timeliness, and coverage of water quality monitoring.

[0020] 2. Pioneering on-site dynamic correction mechanism for satellite inversion models based on measured data to eliminate the influence of meteorological environment: This invention utilizes UAV hyperspectral data and measured water quality parameters from unmanned surface vessels to construct a ground hyperspectral inversion model to obtain a high-precision water quality parameter layer. Then, a scale elimination algorithm is used to solve the scale mismatch problem between satellite and UAV data, and a weighted least squares method is used to iteratively optimize the parameters of the satellite empirical inversion model. This effectively eliminates the interference of the day's meteorological environment conditions on the accuracy of satellite inversion, significantly improves the professionalism and practical value of satellite remote sensing inversion layers, and provides accurate and reliable data support for environmental decision-making.

[0021] 3. Design a precise planning algorithm for sampling areas and collection points to ensure the quality of on-site measured data: Based on the alarm area range, the drone shooting area and the preset coverage ratio, this invention automatically generates the optimal layout scheme of the drone sampling area and the unmanned vessel collection points through steps such as outer rectangular grid division and uniform step size selection. This ensures that the on-site measured data can efficiently, uniformly and representatively cover the abnormal area, providing high-quality basic data for subsequent model correction.

[0022] 4. Forming a closed-loop process of "monitoring-correction-analysis-decision" to enhance emergency response capabilities: This invention constructs a complete closed-loop system from satellite remote sensing anomaly identification, on-site collaborative measurement, model dynamic correction, comprehensive water quality analysis to real-time linkage with the command center, realizing rapid response and precise handling of water quality emergencies, and significantly improving the scientific nature, timeliness and decision support capabilities of environmental management and emergency response. Attached Figure Description

[0023] Figure 1 This is a block diagram of the high-precision collaborative emergency response system of satellite-monitoring vehicle of the present invention.

[0024] Figure 2 This is a flowchart of the high-precision collaborative emergency response method between satellite and monitoring vehicle according to the present invention.

[0025] Figure 3 This is a flowchart of step S4 in the high-precision collaborative emergency response method of satellite-monitoring vehicle of the present invention.

[0026] Figure 4 This is a flowchart of step S5 in the high-precision collaborative emergency response method of satellite-monitoring vehicle.

[0027] Figure 5 This is a schematic diagram of the ground remote sensing reflectance (a) and chlorophyll a band sensitivity analysis (b) in an embodiment of the present invention.

[0028] Figure 6 This is a schematic diagram of a three-band empirical model for chlorophyll a concentration inversion based on long-term historical data, as described in an embodiment of the present invention.

[0029] Figure 7 This is a schematic diagram illustrating the accuracy verification of a random forest model constructed based on UAV hyperspectral data and measured chlorophyll a concentration from an unmanned surface vessel, according to an embodiment of the present invention.

[0030] Figure 8 This is a schematic diagram of the three-band empirical model for chlorophyll a concentration inversion based on the calibration algorithm optimized according to an embodiment of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figure 1 This invention provides a high-precision collaborative emergency response system integrating satellite and monitoring vehicle technologies. The system includes a central control platform integrated into an intelligent remote sensing monitoring vehicle, a satellite remote sensing monitoring system communicatively connected to the central control platform, a drone and unmanned surface vessel (USV) management system, a data processing system, and a decision-making and command system. The core of this invention is to combine satellite remote sensing technology, drone technology, and USV technology. Through system integration and collaborative control, it fully leverages the advantages of each technology to address the pain points of existing water quality monitoring technologies, such as insufficient accuracy, limited coverage, and delayed response. This achieves a comprehensive improvement in the efficiency, accuracy, timeliness, and coverage of water quality monitoring, meeting the dual needs of emergency monitoring of sudden water pollution events and routine water quality monitoring.

[0033] The satellite remote sensing monitoring system has the following core function: to construct an empirical model for water quality parameter inversion and to trigger water quality anomaly alarms. Its working principle is as follows: based on long-term historical ground-measured water quality parameter data of the target water area, combined with synchronous satellite imagery spectral data, an empirical model for water quality parameter inversion is constructed. This model is used to invert and obtain the water quality inversion layer of the entire target water area. Simultaneously, in conjunction with relevant water quality standards, alarm thresholds for each water quality parameter are set. When the water quality parameter in a certain area of ​​the water quality parameter layer inverted by the satellite remote sensing monitoring system reaches or exceeds the alarm threshold, an anomaly alarm is immediately triggered, providing accurate target area information for subsequent emergency monitoring.

[0034] Drone and Unmanned Vessel Management System: This module is the core management unit for on-site measurements. It is responsible for receiving water quality anomaly alarm information from the satellite remote sensing monitoring system and completing the scheduling of monitoring vehicles and the coordinated control of drones and unmanned vessels. Its working principle is as follows: After receiving a water quality anomaly alarm, the central control platform of the intelligent remote sensing monitoring vehicle determines the monitoring area of ​​the drone and the collection point of the unmanned vessel based on the alarm information, and then generates a flight path plan. It automatically dispatches the drone to the abnormal water area. The management system synchronously coordinates the workflow of the drone and the unmanned vessel, automatically planning the drone's shooting area, flight path and the unmanned vessel's water collection point to ensure efficient collaboration between the drone and the unmanned vessel. Finally, it obtains measured hyperspectral data and water quality parameter data from multiple points in the abnormal water area, providing real-time measured basic data for model calibration.

[0035] Data Processing System: This module serves as the system's "data hub," responsible for processing, analyzing, and calibrating all monitoring data. Its core principle is as follows: It receives hyperspectral image data transmitted from UAVs and measured water quality parameter data transmitted from unmanned surface vessels. Using this measured data, it automatically constructs a ground-based hyperspectral inversion model and applies it to the UAV-captured area to invert and obtain high-precision water quality parameter layer data from multiple sampling areas. Subsequently, by combining the inverted high-precision water quality parameter layer data with satellite spectral data, it recalibrates the initial satellite inversion model of the satellite remote sensing monitoring system, correcting the impact of meteorological and environmental factors on the inversion accuracy and improving the accuracy of the satellite inversion layer.

[0036] Decision-making and command system: This module is the "command core" of the system, responsible for building an efficient linkage channel between the field and the command center. Its function is to establish a real-time video conferencing channel between the intelligent remote sensing monitoring vehicle and the command center, realize the real-time transmission of field monitoring data and images, facilitate the command center staff to grasp the field monitoring situation in real time, make accurate decisions and commands for the on-site coordination and dispatch of monitoring vehicles, drones, and unmanned boats, and improve the efficiency of emergency response.

[0037] Please see Figure 2 This invention also provides a high-precision collaborative emergency response method for satellite-monitoring vehicle, which uses the above-mentioned system and includes the following steps.

[0038] Step 1: Based on long-term historical ground-measured water quality parameter data, ground water body remote sensing reflectance data, and synchronous satellite image spectral data of the target water area, conduct water quality parameter optical characteristic analysis and water quality parameter sensitive band analysis, and finally construct an empirical model for water quality parameter inversion.

[0039] Step 1a: Data Collection: The system collects long-term historical data of the target water area, including ground-measured water quality parameters, ground water body remote sensing reflectance data, and satellite image spectral data synchronized with ground measurements, to ensure the integrity, continuity and synchronization of the data, and to provide basic data support for model construction.

[0040] Step 1b: Data preprocessing: Using the satellite hyperspectral response function, the ground water remote sensing reflectance data is converted into equivalent remote sensing reflectance data to eliminate the spectral differences between ground measurements and satellite remote sensing and ensure data compatibility.

[0041] Step 1c: Characteristic Analysis and Sensitive Band Selection: Based on equivalent remote sensing reflectance data, conduct optical characteristic analysis of water quality parameters in the target water area to clarify the material composition and optical characteristic patterns of the water body; at the same time, conduct correlation analysis between various water quality parameters and remote sensing reflectance to select sensitive bands that respond to changes in various water quality parameters, providing targeted parameters for model construction.

[0042] Step 1d: Empirical Model Construction: Based on equivalent remote sensing reflectance data and selected sensitive bands, various empirical models are constructed for the water quality parameters of the target water area, including single-band models, band ratio models, three-band models, etc., to ensure the diversity and applicability of the models.

[0043] Step 1e: Model evaluation and selection: Based on synchronous satellite image spectral data, the accuracy and robustness of various empirical models are evaluated, and the water quality parameter inversion model with the highest accuracy and best robustness is selected as the core model of the satellite remote sensing monitoring system.

[0044] Step 2: Integrate the water quality parameter inversion model selected in Step 1 into the satellite remote sensing monitoring system, and set alarm thresholds and alarm areas in conjunction with relevant water quality standards to achieve automatic alarm for water quality anomalies.

[0045] Step 2a: Threshold setting: Integrate the optimal water quality parameter inversion empirical model into the satellite remote sensing monitoring system, and combine it with local water quality parameter standards, the "Surface Water Environmental Quality Standard" (GB3838-2002) and other relevant specifications to clarify the alarm thresholds for each water quality parameter, and set reasonable alarm area thresholds to ensure the scientific nature and pertinence of the alarms.

[0046] Step 2b: Anomaly Alarm Trigger: The satellite remote sensing monitoring system continuously runs the optimal inversion model to monitor the target water area in real time. When the water quality parameters of a certain area in the water quality parameter layer obtained by inversion reach the set alarm threshold and the area of ​​the abnormal area reaches the alarm area threshold, the system immediately sends alarm information to the central control platform of the intelligent remote sensing monitoring vehicle. The alarm information includes the four-way coordinate range of the abnormal water quality area, the abnormal area, abnormal parameters and other core contents, providing accurate basis for on-site dispatch.

[0047] Step 3: After receiving the alarm information, the central control platform of the intelligent remote sensing monitoring vehicle dispatches the drone and unmanned vessel management system, reasonably delineates the sampling area and flight path, and starts the drone and unmanned vessel to carry out on-site measurement and transmit data back in real time.

[0048] Step 3a, Sampling area and collection point planning: The central control platform of the intelligent remote sensing monitoring vehicle plans the number of sampling areas and collection points of the unmanned vessel reasonably according to the alarm range A and the single shooting range B of the UAV through a special algorithm. The specific contents of the UAV and unmanned vessel monitoring area algorithm are as follows: (1) Input: 1) Alarm range area: A alert(1) Calculate the total area of ​​the rectangle surrounding the alarm area to ensure coverage of all abnormal areas); 2) Area of ​​a single hyperspectral lens shot: B (Determine the shooting area of ​​each hyperspectral image based on the set drone flight altitude to ensure shooting accuracy); 3) Coverage ratio: P (Set the ratio of the alarm area to be covered, for example, P=0.1 means that 10% of the alarm area needs to be covered, which can be adjusted according to the monitoring accuracy requirements).

[0049] (2) Output results: 1) Number of sampling areas N (to ensure coverage of the alarm area of ​​the set proportion); 2) UAV monitoring area C drone (Define the specific coordinate range of each sampling area of ​​the UAV); 3) UAV sampling point P boat (Clarify the specific water sampling coordinates of the unmanned vessel in each sampling area).

[0050] (3) Algorithm steps: 1) Calculate the area of ​​the alarm zone to be covered: A coverag =A alert *P; 2) Calculate the number of sampling areas: Based on the area to be covered and the area captured by the drone in a single shot, calculate the required number of sampling areas N = A coverage / B (round up to ensure coverage of the target area); 3) Determine the side length of a single sampling area: s= (Based on the shooting area B, calculate the square side length of a single sampling area to ensure uniform sampling area specifications); 4) Generate the outer rectangle of the alarm area: Generate the outer rectangle according to the four boundary coordinates of the alarm area, and determine the width W and height H of the outer rectangle; 5) Outer rectangle grid division: To achieve uniform spatial distribution of the sampling area, the outer rectangle is divided into grids, and the number of sampling areas that can be divided in each row and column of the outer rectangle is calculated: n=W / s (number of columns), m=H / s (number of rows); 6) Initialize the potential sampling area position: The potential sampling area position is generated using a grid system, and the coordinates of each potential position (x,y) are: C potential = {(j * s, i * s) | i=0,1,...,m-1; j=0,1,...,n-1}, ensuring that the potential sampling area is uniformly distributed within the bounding rectangle; 7) Uniformly select sampling areas: calculate the uniform selection step size step=(n*m) / N, select sampling areas from the potential sampling area positions according to the step size, and finally determine the UAV monitoring area C. drone = {C potential [k]|k = 0,step, 2*step,...}, ensuring the sampling area uniformly covers the target area; 8) Determine the unmanned surface vessel (USV) sampling point: The USV sampling point for each sampling area is set as the center position of that sampling area, and the coordinate calculation formula is: P boat ,k =(C drone,k,x +(s / 2),C drone,k,y+(s / 2)) (where k=0,1,...,N-1), to ensure that the collection point can represent the water quality of the sampling area.

[0051] Step 3b: Route planning: The UAV and unmanned surface vessel control system outputs the UAV monitoring area C based on the algorithm described above. drone Unmanned boat data collection point P boat The system automatically plans the flight paths of drones (to ensure coverage of all sampling areas) and the navigation paths of unmanned vessels (to ensure arrival at all collection points). The route planning must take into account both monitoring efficiency and data accuracy, and avoid route overlap and omissions.

[0052] Step 3c: On-site data acquisition and transmission: The UAV automatically acquires hyperspectral image data of each sampling area according to the planned route and transmits the acquired image data back to the data processing system in real time; the unmanned surface vessel automatically goes to each collection point according to the planned route, collects water quality parameter data through the onboard water quality instrument, and transmits the measured water quality parameter data back to the data processing system in real time to ensure the real-time performance and integrity of the data.

[0053] Step 4: After receiving the real-time data transmitted back by the UAV and the unmanned vessel, the data processing system constructs a ground hyperspectral inversion model and applies the model to obtain high-precision water quality parameter layer data of the sampling area.

[0054] Step 4a: Inversion Model Optimization and Construction: The data processing system matches the returned UAV hyperspectral data with the measured water quality parameter data of the UAV to construct the model training dataset and test dataset; by iteratively optimizing the band combination method of each empirical model, the model parameters are adjusted, and the ground hyperspectral inversion model with the highest accuracy is selected to ensure that the model can accurately reflect the correspondence between the water quality parameters of the sampling area and the hyperspectral data.

[0055] Step 4b: Water quality parameter inversion in the sampling area: The optimal ground hyperspectral inversion model selected in Step 4a is applied to the hyperspectral image data of each UAV sampling area to obtain high spatial resolution water quality parameter inversion layer data for each sampling area, providing accurate measured comparison data for subsequent satellite inversion model calibration. The specific process is as follows: Figure 3 As shown.

[0056] Step 5: Using a scale elimination algorithm, the sub-meter-level water quality parameter inversion layer data obtained in Step 4b is resampled to the satellite data resolution. Based on the resampled satellite-scale water quality data, combined with the optimal satellite inversion model selected in Step 1, the model parameters are adjusted using a dedicated calibration algorithm to finally obtain the optimal inversion model algorithm for the current satellite data (i.e., the calibrated satellite inversion model). This eliminates the influence of the day's meteorological environment on the satellite inversion accuracy, achieving on-site dynamic calibration of the satellite inversion model. The specific sub-steps are as follows (see...). Figure 4 ).

[0057] Step 5a, Scale Elimination Algorithm – This core algorithm addresses the scale mismatch between high-resolution UAV data and low-resolution satellite data, achieving precise alignment and ensuring that the resampled data accurately reflects the water quality at the satellite scale. 1) Input parameters: Sub-meter level water quality parameter inversion layer data of the sampling area (resolution 0.5-1m), satellite image data resolution (denoted as R, unit: m), sampling area boundary coordinates, and corresponding satellite image area boundary coordinates; 2) Preprocessing: Geometric correction is performed on the sub-meter level water quality parameter inversion layer data of the sampling area to ensure complete alignment of its spatial coordinates with the corresponding area in the satellite image, eliminating spatial offset errors; simultaneously, outliers (such as extreme water quality parameter values ​​caused by monitoring errors) are removed, and the mean-filling method is used to supplement missing data, ensuring data integrity; 3) Scale Conversion: A weighted least squares scale conversion algorithm is used to resample the sub-meter level data to satellite resolution. This algorithm introduces spatial weight factors and spectral similarity weight factors to solve the problems of boundary distortion and detail loss in traditional interpolation algorithms, achieving precise scale matching between high-resolution data and low-resolution satellite data. This study employs a dual-weighting factor synergistic constraint to ensure both the influence of spatial proximity on water quality parameters and the consistency of spectral characteristics. This effectively avoids the distortion problem of traditional interpolation algorithms in regions with large water quality gradients, improves the accuracy of resampled data, and provides more reliable basic data for subsequent satellite inversion model correction. The specific algorithm flow and formulas are as follows.

[0058] ① Definition of weighting factor: Spatial weighting factor Based on sub-meter pixel With satellite pixel center The Euclidean distance is calculated such that closer objects have greater weights, and the formula is as follows: .

[0059] in, The Euclidean distance between the center of the sub-meter pixel and the center of the satellite pixel; This is the spatial bandwidth parameter, which is set to half the satellite pixel side length R. =R / 2, used to control the decay rate of spatial weights.

[0060] Spectral similarity weighting factor The calculation is based on the cosine similarity between the hyperspectral reflectance of sub-meter pixels and the average hyperspectral reflectance of satellite pixels. Higher spectral similarity results in greater weight. The formula is as follows: .

[0061] Where k is the number of hyperspectral bands; Sub-meter pixel Reflectivity in the k-th band; This represents the average reflectance of all sub-meter pixels within the satellite pixel range in the k-band.

[0062] Comprehensive weighting factor The spatial weights and spectral similarity weights are normalized and then weighted and fused, using the following formula: .

[0063] Where n is the total number of sub-meter level pixels contained within the satellite pixel range.

[0064] ② Scale conversion calculation: Water quality parameter values ​​of satellite pixels Water quality parameter values ​​of all sub-meter pixels within its range The core formula is obtained by weighted summation based on comprehensive weights: .

[0065] 4) Output results: Satellite-scale water quality parameter data of the sampling area with the same resolution as satellite data. This data can be directly matched with satellite spectral data for model correction.

[0066] Step 5b, Satellite Inversion Model Correction Algorithm: Based on the scale-reduced satellite-scale water quality data, the optimal satellite inversion model parameters selected in Step 1 are dynamically corrected to improve the model's inversion accuracy under the meteorological conditions of the day. The specific algorithm is as follows: 1) Input parameters: Scale-reduced satellite-scale water quality parameter data (measured calibration data), satellite spectral data of the corresponding region, the optimal satellite inversion model selected in Step 1 (denoted as M0), initial model parameters (denoted as θ0), and correction error threshold (denoted as ε, which can be set according to the monitoring accuracy requirements, usually taken as 0.05); 2) Data matching: The satellite-scale water quality parameter data is matched one-to-one with the satellite spectral data of the corresponding region to construct a model correction dataset, where the satellite spectral data is used as the input variable and the satellite-scale water quality parameter data is used as the true label value; 3) Parameter iterative correction: The initial parameter θ0 of model M0 is iteratively optimized using the weighted least squares (WLS) method. The core is to adjust the model parameters by minimizing the weighted mean square error (WMSE). After each parameter adjustment, the confidence weight is combined with the data. Calculate the weighted mean square error (WMSE) between the model inversion values ​​and the true label values. The optimization objective is to minimize the weighted mean square error between the model inversion values ​​and the true label values. The objective function is... Defined as: ; ; .

[0067] in, Let be the objective function. To correct the total number of samples (satellite pixels) in the dataset; True label value of satellite pixels Confidence weights (composed of comprehensive weights) (obtained through conversion) This is the mathematical expression of the optimal satellite inversion model; Let be the spectral feature vector of the j-th satellite pixel; These are the actual labeled values ​​of water quality parameters at the satellite scale. For drones with sub-meter pixel resolution Water quality parameters The measured confidence level is calculated based on the in-situ measured data of the unmanned vessel and the inverted data of the UAV. For drones with sub-meter pixel resolution Water quality parameter values ​​monitored in situ by the nearest unmanned surface vessel (unit: Ω) Consistent (e.g., mg / L) For drones with sub-meter pixel resolution Water quality parameter inversion values ​​(unit: ) (Consistent), obtained by inversion from the ground hyperspectral inversion model constructed in step 4b.

[0068] If the sub-meter level pixels of a drone within a certain satellite pixel range have high spatial proximity and good spectral consistency (i.e., ... If the overall confidence level is too high, then the transformed confidence weight will be higher. A higher value indicates that the actual label value of the satellite pixel is higher. More reliable, in iterative correction, this The error compared to the model inversion value will be given a higher weight and will have a greater impact on the adjustment of the model parameter θ; if the UAV sub-meter level pixels within a certain satellite pixel range have large spatial dispersion and spectral differences (i.e. If the overall confidence level is too low, then the converted confidence weight will be lower. The lower value indicates that the actual label value of the satellite pixel is too low. The reliability is low (potentially due to issues such as contamination boundary mixing and monitoring errors), and in iterative correction, this... The error between the model inversion value and the actual value will be given a lower weight, reducing the interference on the model parameter θ.

[0069] Gradient descent is used to iteratively update the parameters. The parameters are adjusted along the negative gradient of the objective function to ensure convergence. The parameter update formula for the (k+1)th iteration is: , Let be the parameter vector for the k-th iteration, with the initial value θ0 determined by step 1e, and η be the learning rate; For the objective function in gradient at, These are the partial derivatives of the model with respect to the parameters, the specific form of which is determined by the empirical model type in step 1d: .

[0070] After each iteration, the root mean square residual (RMSE) of the current iteration is calculated. The iteration is terminated when the RMSE is less than a set threshold ε, or when the maximum number of iterations K is reached. max Stop iterating when the time comes. .

[0071] Step 6: Substitute the corrected model parameters (denoted as θ1) into model M0 to obtain the corrected satellite inversion model (denoted as M1). Apply model M1 to the satellite remote sensing data of the day to obtain high-precision water quality parameter data. The data processing system integrates and analyzes the water quality layer data obtained from satellite inversion, UAV hyperspectral data, and UAV water quality meter measured data. Combined with water quality evaluation standards, it conducts comprehensive water quality analysis and automatically generates a standardized water quality report. The report includes core content such as details of water quality parameters, distribution of abnormal areas, and water quality level evaluation. Subsequently, through the real-time linkage channel between the intelligent remote sensing monitoring vehicle and the command center, real-time on-site reporting of water quality emergencies is completed, and water quality reports, on-site images, and various monitoring data are transmitted synchronously. Command center staff conduct a comprehensive evaluation and analysis of the water quality report and make scientific emergency response decisions based on the actual on-site situation, realizing closed-loop management of "monitoring-correction-analysis-decision" and effectively improving the efficiency and scientific nature of water quality emergency response.

[0072] The following is a specific example to illustrate this.

[0073] Step 1, taking Danjiangkou Reservoir as an example, based on long-term historical ground-measured chlorophyll a concentration data and ground water remote sensing reflectance data (such as... Figure 5 Using the spectral data of Landsat 8 OLI satellite imagery (as shown in (a)) and the optical characteristics of chlorophyll a water bodies, as well as the analysis of chlorophyll a sensitive bands, a high-precision empirical model for water quality parameter inversion was finally constructed.

[0074] Based on the conclusions of chlorophyll a sensitive band analysis (such as...) Figure 5 As shown in (b), the Landsat 8 OLI sensor chlorophyll a (Chla) three-band II empirical inversion model constructed by combining empirical model algorithms has the best inversion accuracy (e.g., Figure 6 As shown), the model expression is as follows: .

[0075] in, (1 / OLI2-1 / OLI3)*OLI4, where bands OLI2, OLI3, and OLI4 correspond to the second, third, and fourth bands of Landsat 8 OLI data, respectively. Step 2: Integrate the optimal water quality inversion model from Step 1 into the satellite remote sensing monitoring system, and set the alarm threshold to 15 ug / L and the alarm area to 10 km² according to the "Surface Water Environmental Quality Standard" (GB3838-2002). 2 This enables automatic alarms for abnormal water quality.

[0076] Step 3: After receiving the alarm information, the central control platform of the intelligent remote sensing monitoring vehicle dispatches the drone and unmanned vessel management system, reasonably delineates the sampling area and flight path, and starts the drone and unmanned vessel to carry out on-site measurement and transmit data back in real time.

[0077] Step 4: After receiving real-time data from the UAV and unmanned surface vessel, the data processing system constructs a ground hyperspectral inversion model, as shown below. The random forest model achieves an accuracy of R² as high as 0.94 (e.g., Figure 7 As shown in the figure, the model is applied to obtain high-precision water quality parameter layer data of the sampling area.

[0078] Step 5: Using the scale elimination algorithm, the sub-meter level water quality parameter inversion layer data of the sampling area obtained in Step 4b is resampled to the resolution of satellite data. Based on the resampled satellite-scale water quality data, combined with the optimal satellite inversion model selected in Step 1, the model parameters are adjusted through a dedicated correction algorithm to finally obtain the optimal inversion model algorithm for the current satellite data, eliminating the influence of the meteorological environment on the accuracy of satellite inversion and realizing the on-site dynamic calibration of the satellite inversion model.

[0079] After optimization using the correction algorithm, the model fitting accuracy was significantly improved, with the coefficient of determination R² reaching 0.81 (e.g. Figure 8 (As shown). The optimization results of the model parameters are as follows: original constant term Updated to 2.8594, intercept term changed from The value is corrected to -0.167. The optimal inversion model algorithm for the current satellite data is as follows: .

[0080] in, (1 / OLI2-1 / OLI3)*OLI4, where bands OLI2, OLI3, and OLI4 correspond to the second, third, and fourth bands of Landsat 8 OLI data, respectively.

[0081] Step 6: The data processing system integrates and analyzes the water quality layer data obtained from satellite inversion, UAV hyperspectral data, and unmanned surface vessel water quality meter measurements. It conducts comprehensive water quality analysis based on water quality evaluation standards and automatically generates a standardized water quality report. The report includes core content such as detailed water quality parameters, distribution of abnormal areas, and water quality grade assessment. Subsequently, through the real-time linkage between the intelligent remote sensing monitoring vehicle and the command center, real-time on-site reporting of water quality emergencies is completed, simultaneously transmitting water quality reports, on-site images, and various monitoring data. Command center staff conduct a comprehensive evaluation and analysis of the water quality report, making scientific emergency response decisions based on the actual on-site situation. This achieves closed-loop management of "monitoring-correction-analysis-decision," effectively improving the efficiency and scientific rigor of water quality emergency response.

[0082] This invention has the following features and effects: 1. It constructs a collaborative monitoring system of satellite, UAV, and unmanned vessel, using an intelligent remote sensing monitoring vehicle as the integrated carrier, integrating the advantages of the three to solve the pain points of insufficient accuracy, limited coverage, and delayed response in traditional monitoring; 2. It proposes a dynamic on-site calibration scheme for satellite inversion models based on UAV-unmanned vessel measured data, introducing confidence weights from comprehensive weight transformation, and combining weighted least squares method to optimize parameters, eliminating the influence of the daily meteorological environment on the accuracy of satellite inversion; 3. It designs a precise planning algorithm for UAV sampling areas and unmanned vessel collection points, achieving efficient and accurate collection of on-site measured data, providing a reliable foundation for model calibration; 4. It forms a closed-loop process of "model construction - anomaly alarm - on-site measurement - model calibration - analysis and decision-making," achieving dual adaptation for emergency water quality monitoring and routine monitoring.

[0083] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A high-precision collaborative emergency response system combining satellite and monitoring vehicle, characterized in that: This includes a central control platform integrated into the intelligent remote sensing monitoring vehicle, a satellite remote sensing monitoring system that communicates with the central control platform, a drone and unmanned vessel management and control system, and a data processing system; The satellite remote sensing monitoring system is used to construct an empirical model for water quality parameter inversion based on historical ground-measured water quality parameter data of the target water area, ground water body remote sensing reflectance data, and synchronous satellite image spectral data. The system then uses the empirical model to invert and obtain the water quality inversion layer of the entire target water area. When the water quality parameter of a certain area in the water quality inversion layer reaches or exceeds a preset alarm threshold, an alarm message is generated and sent. The central control platform is used to receive the alarm information, and determine the monitoring area of ​​the UAV and the data collection point of the UAV based on the alarm information, and then generate a route path plan. The drone and unmanned vessel management system is used to control the drone to collect hyperspectral image data of the sampling area according to the route planning, and to transmit the collected hyperspectral image data back to the data processing system in real time; according to the route planning, the unmanned vessel is controlled to reach the preset collection point to conduct in-situ monitoring of water quality parameters, and to transmit the measured water quality parameter data back to the data processing system in real time. The data processing system is used to receive the hyperspectral image data and measured water quality parameter data, construct a ground hyperspectral inversion model, obtain high-precision water quality parameter layer data of the sampling area, and dynamically correct the water quality parameter inversion empirical model based on the high-precision water quality parameter layer data and satellite spectral data to generate a corrected satellite inversion model; apply the corrected satellite inversion model to the satellite remote sensing data of the day to invert high-precision water quality parameter data, and combine it with UAV hyperspectral data and UAV measured data to perform comprehensive water quality analysis and generate a water quality report; The data processing system dynamically corrects the water quality parameter inversion empirical model based on the high-precision water quality parameter layer data and satellite spectral data, generating a corrected satellite inversion model, specifically including: S601. Using the scale elimination algorithm, the high-precision water quality parameter layer data of the acquired sampling area is resampled to the same resolution as the satellite data to obtain satellite-scale water quality parameter data. S602. Match the satellite-scale water quality parameter data with the satellite spectral data of the corresponding region to construct a calibration dataset; S603. Using the weighted least squares method, the satellite-scale water quality parameter data in the calibration dataset is used as the real label to iteratively optimize the parameters of the constructed water quality parameter inversion empirical model until the preset error threshold is met, thus obtaining the calibrated satellite inversion model.

2. The satellite-monitoring vehicle high-precision collaborative emergency system according to claim 1, characterized in that, The central control platform determines the drone monitoring area and the unmanned surface vessel data collection point based on the alarm information, specifically including: S201、acquire the alarm area A alert , the single shooting range area B of the unmanned aerial vehicle and the preset coverage ratio P; S202、According to the alarm area A alert and the preset coverage ratio P, calculate the alarm area A that needs to be covered coverage =A alert *P; S203, Based on the alarm area A that needs to be covered. coverage Given the area B of a single drone shot, determine the number of sampling areas N = ceil(A) coverage / B), where ceil represents rounding up; S204. Determine the side length of a single sampling area based on the area B of the drone's single-shot coverage. ; S205. Based on the four boundary coordinates of the alarm area, generate an outer rectangle and determine the width W and height H of the outer rectangle; S206. Divide the outer rectangle into grids according to the side length s, and calculate the number of sampling areas that can be divided in each row n = W / s and the number of sampling areas that can be divided in each column m = H / s. S207. Generate the potential sampling area location set C using a grid system. potential = {(j * s, i * s) | i=0,1,...,m-1; j=0,1,...,n-1}, ensuring that the potential sampling area is uniformly distributed within the bounding rectangle; S208. Based on the number of sampling areas N and the total number of potential locations n*m, calculate the uniform selection step size step = (n*m) / N, and uniformly select indices from the set of potential sampling area locations according to the step size step to obtain the UAV monitoring area set C. drone ={C potential [k]|k = 0,step, 2*step,...}; S209. For each selected sampling area, determine its center location as the unmanned surface vessel (USV) sampling point P. boat ,k =(C drone,k,x +(s / 2),C drone,k,y +(s / 2)), where k = 0,1,…,N-1.

3. The satellite-monitoring vehicle high-precision collaborative emergency system according to claim 1, characterized in that, The scale elimination algorithm includes: Spatial weighting factor calculated based on the Euclidean distance between the centers of sub-meter level pixels and satellite pixels. ; Spectral similarity weighting factor is calculated based on the cosine similarity between sub-meter pixel hyperspectral reflectance and average satellite pixel hyperspectral reflectance. ; Spatial weighting factor Weighting factor for spectral similarity After normalization and weighted fusion, a comprehensive weight factor is obtained. ; Based on comprehensive weighting factor The water quality parameter values ​​of the satellite pixels are obtained by weighted summation of the sub-meter level pixel values. .

4. The satellite-monitoring vehicle high-precision collaborative emergency system according to claim 3, characterized in that, The objective function of the weighted least squares method is: ； ； ； in, Let be the objective function. To correct the total number of samples in the dataset; True label value of satellite pixels Confidence weights; This is the mathematical expression of the optimal satellite inversion model; Let be the spectral feature vector of the j-th satellite pixel; These are the actual labeled values ​​of water quality parameters at the satellite scale. For drones with sub-meter pixel resolution Water quality parameters The measured confidence level is calculated based on the in-situ measured data of the unmanned vessel and the inverted data of the UAV. For drones with sub-meter pixel resolution The water quality parameters were monitored in situ by the unmanned surface vessel that was closest to the water in the water. For drones with sub-meter pixel resolution The water quality parameters were retrieved from the ground-based hyperspectral retrieval model.

5. A high-precision collaborative emergency response method using satellite and monitoring vehicle, characterized in that, The method, performed using the system described in any one of claims 1-4, comprises the following steps: S1. The satellite remote sensing monitoring system constructs an empirical model for water quality parameter inversion based on historical ground-measured water quality parameter data, ground water body remote sensing reflectance data, and synchronous satellite image spectral data of the target water area. The system inverts and obtains the water quality inversion layer of the entire target water area through the empirical model. When the water quality parameter of a certain area in the water quality inversion layer reaches or exceeds the preset alarm threshold, an alarm message is generated and sent. S2. The central control platform of the intelligent remote sensing monitoring vehicle receives the alarm information, determines the monitoring area of ​​the drone and the collection point of the unmanned vessel based on the alarm area range and the single shooting range of the drone, and then generates a route path plan, which is sent to the drone and unmanned vessel management systems respectively. S3. The UAV and unmanned vessel management system, according to the route path planning, controls the UAV to collect hyperspectral image data of the sampling area and transmits the collected hyperspectral image data back to the data processing system in real time. It also controls the unmanned vessel to reach the preset collection point to conduct in-situ monitoring of water quality parameters and transmits the measured water quality parameter data back to the data processing system in real time. S4. The data processing system receives the hyperspectral image data and measured water quality parameter data, constructs a ground hyperspectral inversion model, and obtains high-precision water quality parameter layer data of the sampling area. S5. Based on the high-precision water quality parameter layer data and satellite spectral data of the sampling area obtained in step S4, the data processing system dynamically corrects the water quality parameter inversion empirical model constructed in step S1 and generates the corrected satellite inversion model. S6. Apply the corrected satellite inversion model to the satellite remote sensing data of the day to obtain high-precision water quality parameter data. Combine this with UAV hyperspectral data and UAV measured data to conduct a comprehensive water quality analysis and generate a water quality report.

6. The high-precision collaborative emergency response method between satellite and monitoring vehicle according to claim 5, characterized in that, Step S2, which involves determining the UAV monitoring area and the UAV data collection point based on the alarm information, specifically includes: S201, Obtain the alarm area area A alert The area B of a single drone shot and the preset coverage ratio P; S202, Based on the alarm area A alert Calculate the alarm area A to be covered based on the preset coverage ratio P. coverage =A alert *P; S203, Based on the alarm area A that needs to be covered. coverage Given the area B of a single drone shot, determine the number of sampling areas N = ceil(A) coverage / B), where ceil represents rounding up; S204. Determine the side length of a single sampling area based on the area B of the drone's single-shot coverage. ; S205. Based on the four boundary coordinates of the alarm area, generate an outer rectangle and determine the width W and height H of the outer rectangle; S206. Divide the outer rectangle into grids according to the side length s, and calculate the number of sampling areas that can be divided in each row n = W / s and the number of sampling areas that can be divided in each column m = H / s. S207. Generate the potential sampling area location set C using a grid system. potential = {(j * s, i * s) | i=0,1,...,m-1; j=0,1,...,n-1}, ensuring that the potential sampling area is uniformly distributed within the bounding rectangle; S208. Based on the number of sampling areas N and the total number of potential locations n*m, calculate the uniform selection step size step = (n*m) / N, and uniformly select indices from the set of potential sampling area locations according to the step size step to obtain the UAV monitoring area set C. drone ={C potential [k]|k = 0,step, 2*step,...}; S209. For each selected sampling area, determine its center location as the unmanned surface vessel (USV) sampling point P. boat ,k =(C drone,k,x +(s / 2),C drone,k,y +(s / 2)), where k = 0,1,…,N-1.

7. The high-precision collaborative emergency response method between satellite and monitoring vehicle according to claim 6, characterized in that, Step S6 specifically includes: S601. Using the scale elimination algorithm, the high-precision water quality parameter layer data of the sampling area obtained in step S5 is resampled to the same resolution as the satellite data to obtain satellite-scale water quality parameter data. S602. Match the satellite-scale water quality parameter data with the satellite spectral data of the corresponding region to construct a calibration dataset; S603. Using the weighted least squares method, with the satellite-scale water quality parameter data in the calibration dataset as the true label, the water quality parameter inversion empirical model constructed in step S1 is iteratively optimized until the preset error threshold is met, and the corrected satellite inversion model is obtained.

8. The high-precision collaborative emergency response method between satellite and monitoring vehicle according to claim 7, characterized in that, The scale elimination algorithm described in step S601 includes: Spatial weighting factor calculated based on the Euclidean distance between the centers of sub-meter level pixels and satellite pixels. ; Spectral similarity weighting factor is calculated based on the cosine similarity between sub-meter pixel hyperspectral reflectance and average satellite pixel hyperspectral reflectance. ; Spatial weighting factor Weighting factor for spectral similarity After normalization and weighted fusion, a comprehensive weight factor is obtained. ; Based on comprehensive weighting factor The water quality parameter values ​​of the satellite pixels are obtained by weighted summation of the sub-meter level pixel values. .

9. The high-precision collaborative emergency response method between satellite and monitoring vehicle according to claim 8, characterized in that, The objective function of the weighted least squares method described in step S603 is: ； ； ； in, Let be the objective function. To correct the total number of samples in the dataset; True label value of satellite pixels Confidence weights; This is the mathematical expression of the optimal satellite inversion model; Let be the spectral feature vector of the j-th satellite pixel; These are the actual labeled values ​​of water quality parameters at the satellite scale. For drones with sub-meter pixel resolution Water quality parameters The measured confidence level is calculated based on the in-situ measured data of the unmanned vessel and the inverted data of the UAV. For drones with sub-meter pixel resolution The water quality parameters were monitored in situ by the unmanned surface vessel that was closest to the water in the water. For drones with sub-meter pixel resolution The water quality parameters were retrieved from the ground-based hyperspectral retrieval model.

Images